ZOZO Blames the Sky, Then Quietly Repositions Around AI Agents

A soft summer, an El Niño winter warning, and a CEO asking the only question that matters in retail: who owns the checkout when the buyer is a bot?

3092.T · Earnings Call · 2026-10-29

The Weather Alibi

ZOZO's fiscal Q2 arrived as a meteorological report with a P&L stapled to it. The summer was mild where last year was scorching, so July and August comped badly; September turned cold, and heavy knits sold. COO Fuminori Hirose laid out the ranking bluntly: “Last year, as you mentioned, last year was hot. This year was more moderate. And so, July, August, Y-o-Y wasn't that great this fiscal year.” — Fuminori Hirose, Director and COO · 2026-10-29 Then the kicker — “September was great this year. September was the best and then July and August was -- if you rank the growth rate, that would be the ranking.” — Fuminori Hirose, Director and COO · 2026-10-29 It is a tidy explanation, and General Manager Yusaku Kobayashi pushed it further, arguing that averaged across the quarter the external backdrop was roughly last year's. But the alibi has a shelf life measured in weeks. October came in warm, and Hirose did not pretend otherwise: “October, it was much warmer than last year. That's an external factor. We can't really sit back and relax yet.” — Fuminori Hirose, Director and COO · 2026-10-29 Worse, the winter itself is the problem. Kobayashi on the fall/winter environment: “With El Nino, we're going to probably see a warmer winter this year. It's not a great environment for us. That's a lot of headwinds.” — Yusaku Kobayashi, General Manager of Corporate Planning Office · 2026-10-29 Management's hedge is that brands have spent four or five years merchandising for warm winters and now carry appropriate inventory. That may be true — but it is also an admission that the second half hinges on the weather rather than on ZOZO. On price, the answer was reassuring rather than exciting: average retail price is up only a few percentage points, and Sawada explicitly does not expect demand to crack. That is a notably softer inflation signal than the brand-led price escalation the analyst was probing for.

From Defense to Offense

The most revealing exchange was Yamaoka's framing of the company's AI strategy as historically defensive — protecting platform value — and asking what an offensive version looks like. Hirose's answer was not about new monetization but about conversion physics: the offline niaulab showroom, where stylists dress members in person, sees purchase value double afterward. The bet is that accurate recommendations — algorithmic rather than human — can replicate that lift at scale. The logic is simple: if the recommendation is right, the basket grows. But it also quietly redefines ZOZO from a mall into a styling engine, which is a far more capital- and compute-intensive business than lease-free marketplace economics. That shift is already visible in the cost line. The jump in Other SG&A was attributed to two things: packaging materials from June, and “AWS and cloud server usage rate has also increased as well.” — Yusaku Kobayashi, General Manager of Corporate Planning Office · 2026-10-29 Cloud intensity is the tell — the AI recommendation layer costs real money before it earns any. CFO Koji Yanagisawa cushioned this by saying the second half holds no further cost surprises beyond heavier advertising, but the direction of travel is clear.

Who Owns the Checkout?

Sawada's answer on payments is the single most forward-looking thing said on the call, and it places tiny ZOZO squarely inside the market's loudest current theme. Asked about US payment rails and AI, he reframed it as a structural fork:

from the EC -- with respect to payments, who should make the payment? Is it EC or agent? And of course, there's a payment collection impact, who's going to pay -- which will be better? Is it better if the EC players or the agents do it?
Kotaro Sawada, CEO · 2026-10-29

He then explained why it matters commercially: purchase data is the moat. “Payment data, purchase data is very important. It will be a huge gap whether we have data about what they bought today, 2 days ago.” — Kotaro Sawada, CEO · 2026-10-29 This is exactly the EC players-versus-agent contest now being narrated across the global tape — Adyen keyed Agentic Commerce this week, Twilio talked up its consumer agent line, DoorDash and Pinterest described consumer-facing agent experiences. ZOZO is not leading that wave, but it is one of the few retailers openly naming the disintermediation risk: an agent that picks the outfit could also take the wallet, and with it the data that makes the picking good in the first place.

The Side Dishes Deserve More Attention Than They Got

Three smaller threads are worth flagging. First, the plan miss at ZOZOTOWN was offset by strength at LINE Yahoo, and the reason was promotional cadence: “LINE Yahoo, LY Commerce, they actually had a lot more ZOZO Festivals. The frequency increased. Therefore, this wasn't included in the plan.” — Fuminori Hirose, Director and COO · 2026-10-29 Sawada confirmed the ZOZO Festival calendar is decided jointly and flexibly — meaning a sister platform's marketing appetite can swing ZOZO's own quarterly mix. Second, ZOZOUSED is growing faster than the core mall, and Sawada's explanation is a genuine structural edge: because ZOZO sold the item new, it knows the original price and can price resale without a costly inspection apparatus. Third, the Pokemon T shirt collaboration is selling far beyond plan and ships into Q3 — but the inventory build it caused briefly puzzled management itself, with Kobayashi conceding mid-answer, “That's -- I have forgotten about that.” — Yusaku Kobayashi, General Manager of Corporate Planning Office · 2026-10-29 Lumpy IP hits and weather are, uncomfortably, the same kind of revenue. Also worth noting: Cross border got a real experiment in the MUSINSA pop-up, framed honestly as a proof of concept rather than a sales event, and Sawada spent meaningful airtime on impact from security, calling cyber risk “a urgent matter” — Kotaro Sawada, CEO · 2026-10-29 globally while citing group company ASKUL's earlier incident as the reason for prior investment. No second-half cost was guided for it — a small reassurance in a quarter where cloud spend was already rising. The read: nothing broke at ZOZO this quarter, and nothing dramatically improved either. What changed is the framing. A company that used to describe AI as a shield now describes it as a sales lever, while simultaneously acknowledging that the agent layer could capture the transaction — and the data — before ZOZO ever sees the customer. The warm winter is a forecastable headwind; the checkout question is not.